3. Results

3.1. Clinical Information

Detailed phenotypic and clinical information for all study participants is provided (Table 1). The cohort consisted of 10 male probands (66.7%) and 5 female probands. Four probands (26.7%) were diagnosed with hypoplastic left heart syndrome (HLHS), two of which were also diagnosed with a NDD condition. The most common form of NDD amongst the probands was ADHD, with 13 of 15 diagnosed. Other forms of NDD included autism spectrum disorder (ASD), learning disabilities, obsessive‐compulsive disorder (OCD) and oppositional defiant disorder. Eight probands had at least one parent with at least one diagnosed form of NDD, and of these probands, a further six had an extended family history of NDD.

Family | Individual ID | Relationship | Sex | Cardiac phenotype | Neurodevelopmental disorder | Psychiatric diagnosis | Genetically inferred ethnicity | Extended family history
1 | 1.001 | Proband | M | HLHS | ADHD | Anxiety | AMR | No
1 | 1.002 | Mother | F |  |  | Depression | AMR | No
1 | 1.003 | Father | M |  |  | Depression | AMR | No
2 | 2.001 | Proband | F | HLHS |  |  | AMR | No
2 | 2.002 | Mother | F |  |  |  | AMR | No
2 | 2.003 | Father | M |  |  |  | AMR | No
3 | 3.001 | Proband | M |  | ADHD |  | EUR | Yes
3 | 3.002 | Mother | F |  |  | Depression, PTSD | EUR | Yes
3 | 3.003 | Father | M |  |  | Depression | EUR | Yes
4 | 4.001 | Proband | M |  | ADHD, specific learning disability in reading | Anxiety | EUR | No
4 | 4.002 | Mother | F |  | ADHD |  | AMR | No
4 | 4.003 | Father | M |  | ADHD |  | EUR | No
5 | 5.001 | Proband | F |  | ADHD, oppositional defiant disorder, dyslexia |  | EUR | No
5 | 5.002 | Mother | F |  | ADHD | Anxiety | EUR | No
5 | 5.003 | Father | M |  | ADHD | Anxiety | EUR | No
6 | 6.001 | Proband | M |  | ADHD, intellectual disability |  | AFR | No
6 | 6.002 | Mother | F |  | ADHD |  | AFR | No
6 | 6.003 | Father | M |  | ADHD |  | AFR | No
8 | 8.001 | Proband | M |  | ADHD, intellectual disability |  | AFR | Yes
8 | 8.002 | Mother | F |  |  |  | EUR | Yes
9 | 9.001 | Proband | M |  | ADHD, intellectual disability |  | EUR | No
9 | 9.002 | Mother | F |  |  |  | EUR | No
9 | 9.003 | Father | M |  | ADHD | Depression | EUR | No
11 | 11.001 | Proband | M |  | ADHD |  | Unknown | No
11 | 11.002 | Mother | F |  | ADHD, OCD | Anxiety | Unknown | No
11 | 11.003 | Father | M |  |  |  | Unknown | No
12 | 12.001 | Proband | F |  | ADHD, language disorder, learning disability |  | EUR | Yes
12 | 12.002 | Mother | F |  |  |  | EUR | Yes
12 | 12.003 | Father | M |  | Learning disabilities |  | EUR | No
13 | 13.001 | Proband | M |  | ADHD, OCD, Tourette syndrome |  | AFR | No
13 | 13.002 | Mother | F |  |  |  | EUR | No
13 | 13.003 | Father | M |  | ADHD |  | AFR | No
15 | 15.001 | Proband | M |  | ADHD, learning disability | Generalised anxiety disorder | EUR | Yes
15 | 15.002 | Mother | F |  |  |  | EUR | Yes
15 | 15.003 | Father | M |  |  |  | EUR | Yes
16 | 16.001 | Proband | M | HLHS | ADHD | Anxiety | EUR | No
16 | 16.002 | Mother | F |  |  | Anxiety | EUR | No
16 | 16.003 | Father | M |  | ADHD |  | EUR | No
19 | 19.001 | Proband | F |  | ADHD, ODD, nonverbal learning disorder |  | EUR | Yes
19 | 19.002 | Mother | F |  |  |  | EUR | Yes
19 | 19.003 | Father | M |  |  |  | EUR | No
20 | 20.001 | Proband | F | HLHS |  | Major depressive disorder, Generalised anxiety disorder | EUR | Yes
20 | 20.002 | Mother | F |  |  |  | EUR | Yes
20 | 20.003 | Father | M |  |  |  | EUR | Yes

3.2. WGS Variants

A total of 33 rare potentially deleterious variants and one large deletion were identified across the 14 probands included in this study, in genes known to be associated with either CHD and/or NDD (Tables 2, 3 and 4). We identified three pathogenic (P) and five likely pathogenic (LP) variants/deletions according to ACMG criteria. Of these, three variants/deletions were considered to have a likely impact on the probands′ phenotypes, occurring in autosomal dominant or de novo acting genes (Table 2), and five represented carrier findings in autosomal recessive disease genes (Table 3).

Patient ID | Gene | Genomic (hg38) | Protein/transcript | Inheritance | gnomAD AF (v4.1) | ACMG
9.001 | ∗, ALG3 ∗, CLCN2 ∗, CHRDL1 ∗, ECE2 | NC_000003.12:g.183316907_185835082del |  | De novo | E = not found | P
G = not found
16.001 | MYH6 ∗ | NC_000014.9 : g.23396814A > C | NM_002470.4 : c.5658 + 3_5658 + 6del | Paternal | E = not found | LP
G = not found
19.001 | ARID1B ∗ ^ | NC_000006.12 : g.157198907G > A | NM_001374828.1 : c.4479G > A | De novo | E = fails QC | LP
G = not found

Patient ID | Gene | Genomic (hg38) | Protein/transcript | Inheritance | gnomAD AF (v4.1) | ACMG
2.001 | MYO7A ∗ | NC_000011.10 : g.77207295G > T | NP_000251.3:p.(E1917 ∗) | Maternal | E = 0.00000206 | P
G = not found
3.001 | TWIST2 ∗ | NC_000002.12 : g.238848457C > A | NP_001258822.1:p.(S81 ∗) | Paternal | E = 0 | LP
G = not found
5.001 | GPR179 ∗ | NC_000017.11 : g.38328525C > A | NP_001004334.3:p.(G1682 ∗) | Paternal | E = 0.0000103 | LP
G = 0.0000198
15.001 | COL18A1 ∗ ^ | NC_000021.9 : g.45509603 T > C | NM_001379500.1 : c.3495 + 2 T > C | Paternal | E = 0.00000151 | LP
G = not found
20.001 | ACADM ∗ | NC_000001.11 : g.75732896 T > C | NP_000007.1:p.(M87T) | Maternal | E = not found | P
G = not found

Patient ID | Gene | Genomic (hg38) | Protein/transcript | Inheritance | gnomAD AF (v4.1) | ACMG
1.001 | BCAR1 | NC_000016.10 : g.75229755 T > C | NP_055382.2:p.(H790R) | Paternal | E = 0.000000684 | VUS
G = not found
1.001 | SETX ∗ | NC_000009.12 : g.132298174A > G | NP_055861.3:p.(L1896S) | Maternal | E = 0.00000616 | VUS
G = 0.00000657
2.001 | PAX6 | NC_000011.10 : g.31801587C > A | NP_001355823.1:p.(V125F) | De novo | E = 0.00000137 | VUS
G = 0
3.001 | DNMT3B ∗ | NC_000020.11 : g.32800234G > A | NP_008823.1:p.(G614E) | Maternal | E = 0.00000137 | VUS
G = 0.00000657
4.001 | NEFH | NC_000022.11:g.29490322_29490342del | NP_066554.2:p.(E895_K901del) | Maternal | E = not found | VUS
G = not found
5.001 | PANK2 | NC_000020.11 : g.3910812A > G | NP_001373322.1:p.(K296R) | Paternal | E = 0.0000951 | VUS
G = 0.000092
5.001 | MYO7A ∗ | NC_000011.10 : g.77174781G > A | NP_000251.3:p.(R654H) | Paternal | E = 0.0000207 | VUS
G = 0.0000394
5.001 | VPS13B ∗ | NC_000008.11 : g.99170112C > A | NP_689777.3:p.(P761H) | Maternal | E = 0.00019 | VUS
G = 0.0000987
5.001 | GRM6 ∗ | NC_000005.10 : g.178986615G > A | NP_000834.2:p.(R547C) | Maternal | E = 0.000271 | VUS
G = 0.000249
6.001 | TYR ∗ | NC_000011.10 : g.89178215A > T | NP_000363.1:p.(T88S) | Paternal | E = 0.00000137 | VUS
G = 0.0000329
6.001 | MARCHF8 | NC_000010.11:g.45464286_45464288del | NP_001269795.1:p.(S66del) | Maternal | E = 0.00000547 | VUS
G = 0.00000657
12.001 | MYH3 ∗ | NC_000017.11:g.10629838_10629841del | NC_000017.11(NM_002470.4): c.5658 + 3_5658 + 6del | Maternal | E = 0.000000684 | VUS
G = not found
13.001 | NIPBL ∗ | NC_000005.10 : g.37049221C > T | NP_597677.2:p.(H2292Y) | Paternal | E = not found | VUS
G = not found
13.001 | CLCN7 ∗ | NC_000016.10 : g.1448390C > T | NP_001278.1:p.(G660S) | Paternal | E = 0.0000123 | VUS
G = 0.0000394
13.001 | TWIST1 ∗ | NC_000007.14 : g.19116925 T > G | NP_000465.1:p.(K133Q) | Paternal | E = 0.000000684 | VUS
G = not found
15.001 | PCDH19 ∗ | NC_000023.11 : g.100296433A > C | NP_001171809.1:p.(Y1097 ∗) | Maternal | E = not found | VUS
G = not found
16.001 | RERE ∗ | NC_000001.11 : g.8355435G > A | NP_001036146.1:p.(Q1551 ∗) | Paternal | E = not found | VUS
G = not found
16.001 | OSGEP ∗ | NC_000014.9 : g.20449302C > A | NC_000014.9(NM_017807.4): c.412 − 36G > T | Paternal | E = fails QC | VUS
G = not found
19.001 | PLOD1 ∗ | NC_000001.11 : g.11957873C > T | NP_000293.2:p.(P258L) | Paternal | E = 0.00000821 | VUS
G = 0.000046
19.001 | CAD ∗ | NC_000002.12 : g.27243441G > A | NP_004332.2:p.(A2201T) | Paternal | E = 0.0000575 | VUS
G = 0.0000649
20.001 | JAG1 ∗ | NC_000020.11 : g.10672949C > A | NP_000205.1:p.(G47W) | Paternal | E = 0 | VUS
G = not found
20.001 | AFG3L2 ∗ | NC_000018.10 : g.12359961G > A | NP_006787.2:p.(R240W) | Paternal | E = 0.0000335 | VUS
G = 0.0000131
20.001 | SLC46A1 | NC_000017.11 : g.28405167C > T | NP_542400.2:p.(R177H) | Paternal | E = 0.00000274 | VUS
G = 0.0000263
20.001 | FOXN1 ∗ | NC_000017.11 : g.28535093G > A | NP_001356298.1:p.(E508K) | Paternal | E = 0.0000351 | VUS
G = 0.0000263
20.001 | CNTNAP2 ∗ | NC_000007.14 : g.147486030C > A | NP_054860.1:p.(T589N) | Paternal | E = 0.0000089 | VUS
G = 0.0000131
20.001 | FAM126A ∗ | NC_000007.14 : g.22961300C > G | NP_115970.2:p.(A256P) | Maternal | E = 0.000903 | VUS
G = 0.000671

Amongst the variants with potential phenotypic relevance (Table 2), Proband 9 carried a de novo 2.5‐MB deletion on Chromosome 3 (NC_000003.12:g.183316907_185835082del), encompassing four developmental genes (DVL3, ALG3, CLCN2, ECE2). This deletion is likely to have contributed to the severe phenotype observed in Proband 9, who presented with ADHD, intellectual disability and microcephaly, consistent with haploinsufficiency of multiple neurodevelopmental genes. Of particular interest for CHD‐NDD overlap, Proband 16 (HLHS and ADHD) carried a paternally inherited stop‐gain variant in MYH6 (NC_000014.9 : g.23396814A > C; NP_002462.2:p.Y724∗). In a large genomic study of 11,555 CHD probands, 4% of individuals with a MYH6 variant carried an additional NDD diagnosis [32]; however, given that loss‐of‐function variants in MYH6 are well‐tolerated in the general population (pLI = 0), this co‐occurrence likely reflects the broader CHD‐NDD comorbidity rather than a direct effect of MYH6 on neurodevelopment. The same proband also carried a paternally inherited stop‐gain variant in RERE. This is a variant of uncertain significance, as it is located close to the end of the penultimate exon and is predicted to escape nonsense‐mediated decay, resulting in deletion of less than 1% of the protein. Pathogenic variants in RERE cause a syndromic form of intellectual disability in which ADHD and septal defects are recognised features [33]; it is therefore possible that this VUS may be relevant to this proband′s phenotype, although a contribution cannot be confirmed without further evidence. Finally, Proband 19 carried a de novo splice‐site variant in the chromatin remodelling gene ARID1B (NM_001374828.1 : c.4479G > A), an autosomal dominant gene in which de novo variants cause Coffin–Siris syndrome and intellectual disability [34]. Proband 19 had been diagnosed with ADHD and learning disabilities, consistent with this established disease mechanism.

Five further probands carried single heterozygous pathogenic or LP variants in autosomal recessive disease genes, representing carrier findings rather than direct disease‐causing genotypes (Table 3). Proband 2 with HLHS carried a maternal MYO7A stop‐gain variant (NC_000011.10 : g.77207295G > T; NP_000251.3:p.E1917∗), a gene associated with Usher syndrome type 1B and nonsyndromic deafness. ⁠Proband 3 carried a paternally inherited stop‐gain variant in TWIST2 (NC_000002.12 : g.238848457C > A; NP_001258822.1:p.S81∗). TWIST2 is associated with both autosomal dominant conditions (ablepharon‐macrostomia and Barber–Say syndromes), caused by missense variants and autosomal recessive Setleis syndrome, caused by biallelic loss‐of‐function variants [35]. As heterozygous loss‐of‐function variants in TWIST2 are consistent with carrier status for Setleis syndrome rather than a dominant pathogenic mechanism, this variant was not considered causal for the proband′s phenotype.

Proband 5 carried a paternal GPR179 stop‐gain variant (NC_000017.11 : g.38328525C > A; NP_001004334.3:p.G1682∗), primarily linked to congenital stationary night blindness, though emerging evidence suggests broader neurological involvement in some patients. Proband 15 carried a paternal COL18A1 splice donor variant (NM_001379500.1 : c.3495 + 2 T > C; NC_000021.9 : g.45509603 T > C), associated with Knobloch syndrome, in which cognitive impairment and developmental delays are reported in the majority of cases. Proband 20 with HLHS carried a maternal pathogenic ACADM variant associated with medium‐chain acyl‐CoA dehydrogenase deficiency, a recessive condition which can present with developmental delays secondary to metabolic dysfunction [36]. These variants, found in autosomal recessive genes, indicate carrier status: A second pathogenic variant is required for disease manifestation. At present, all four genes are classified as AR; this may be re‐evaluated as new evidence emerges. Reporting these findings is important for genetic counselling and family planning.

The remaining 26 variants of uncertain significance (Table 4) included genes with strong biological plausibility, potentially warranting future functional studies.

Genome‐wide analysis of loss‐of‐function variants identified six variants in genes intolerant to loss of function (LOEUF ≤ 0.35) (Table 5). Two were previously identified (PCDH19 and RERE), whereas four novel variants were found in CHD9, CNTFR, INTS6 and SPTBN1. Three of these genes have established NDD associations. CNTFR variants have been linked to ADHD severity in multiple studies [37, 38], consistent with the ADHD phenotype in Proband 5. SPTBN1 loss‐of‐function variants cause a spectrum from isolated ADHD to severe intellectual disability [39]; our patient presented with ADHD and intellectual disability, consistent with a potential pathogenic role. INTS6 is a strong ASD candidate gene (SFARI database) and may have broader NDD associations, as suggested by the intellectual disability in Proband 8. The CHD9 variant in Proband 6 lacks specific disease associations but warrants investigation given its chromatin remodelling function, consistent with other pathogenic variants in our cohort.

Patient ID | Gene | Genomic (hg38) | Protein | Inheritance | gnomAD AF | LOEUF
5.001 | CNTFR | NC_000009.12:g.34557589del | NP_671693.1:p.(I181 ∗) | Maternal | E = 0.00000274 | 0.339
G = 0.00000657
6.001 | SPTBN1 | NC_000002.12:g.54668402_54668403insCA | NP_003119.2:p.(K2310Tfs ∗113) | Maternal | E = not found | 0.088
G = not found
6.001 | CHD9 | NC_000016.10:g.53209725_53209728del | NP_001295248.1:p.(K566Vfs ∗4) | Paternal | E = not found | 0.17
G = not found
8.001 | INTS6 | NC_000013.11:g.51395410_51395411del | NP_036273.1:p.(F169Cfs ∗24) | De novo | E = 0.000000684 | 0.136
G = not found
15.001 | PCDH19 ∗ | NC_000023.11 : g.100296433A > C | NP_001171809.1:p.(Y1097 ∗) | Maternal | E = not found | 0.126
G = not found
16.001 | RERE ∗ | NC_000001.11 : g.8355435G > A | NP_001036146.1:p.(Q1551 ∗) | Paternal | E = not found | 0.12
G = not found

3.3. Polygenic Risk Score Analysis

The cohort showed a trend towards higher PRS for ADHD and ASD compared with controls; however, due to the small sample size (n = 9), no score reached statistical significance (Figure S2). Despite this, individual‐level observations support a role for polygenic inheritance in shaping NDD risk, including in the context of identified rare variants.

Patient 4‐001 (NDD) had a PRS in the 89th percentile for ADHD and 83rd percentile for autism, with clinical diagnoses of ADHD and learning disability, and no LP/P monogenic variant identified. This represents the clearest case of a likely polygenic contribution to NDD phenotype in this cohort.

Patient 16‐001 (NDD + CHD) carries a paternally inherited MYH6 truncating variant that is likely relevant to the cardiac phenotype. This patient also had an anxiety PRS in the 88th percentile and a clinical diagnosis of anxiety. Although the contribution of each genetic factor cannot be determined, this observation illustrates how common variant burden may coexist with rare variants in individuals with complex phenotypes.

Similarly, Patient 15‐001 (NDD) carries a paternally inherited LP variant in COL18A1 (NM_001379500.1 : c.3495 + 2 T > C) in a gene associated with autosomal recessive disease, rendering the heterozygous finding unlikely to be fully explanatory. This patient scored in the 97th percentile for anxiety PRS with a corresponding anxiety diagnosis, again suggesting a polygenic contribution to the NDD phenotype.

Although formal epistasis testing is precluded by cohort size, these observations collectively suggest that polygenic burden may act additively with, or independently of, rare variant findings to shape neuropsychiatric outcomes in this cohort, a hypothesis warranting investigation in larger studies.

3.4. Gene Expression

There were 14 genes across three probands that were found to be aberrantly expressed (Table S3). Eight of these genes (EIF4G1, EIF2B5, DVL3, AP2M1, ABCF3, SENP2, PSMD and PARL) were downregulated in Proband 9, who was previously identified as having a large deletion on Chromosome 3, confirming the downstream effect of this finding. We investigated whether genetic variants could explain the other expression changes and identified that Proband 8 had one de novo variant of uncertain significance in PPID (NM_005038.3:c.737_738del), which may have contributed to aberrant PPID expression. No other variants of interest were identified in the remaining aberrantly expressed genes; however, these genes are involved in broad cellular processes that may influence pathways associated with NDD or CHD.

3.5. RNA Sequencing Validates Pathogenic Splicing Defects

Splicing variant validation was performed for three variants predicted to disrupt splice sites. Two showed definitive aberrant splicing. The ARID1B variant (NM_001374828.1 : c.4479G > A) in Proband 19 disrupts the splice donor consensus sequence, resulting in exon skipping absent in both parents, consistent with a de novo origin (Figure S3A). The paternally inherited COL18A1 variant (NM_001379500.1 : c.3495 + 2 T > C) in Proband 15 disrupts the canonical splice donor site, producing two aberrant outcomes (cryptic donor activation with partial intron retention and full intron retention) both introducing a premature stop codon and predicting loss of function (Figure S3B). The MYH3 variant (NM_002470.4 : c.5658 + 3_5658 + 6del) showed suggestive splicing changes, but junction coverage was below the threshold for confident validation (<20 reads). As detailed in the WGS analysis (Tables 2 and 3), variants in autosomal recessive genes (COL18A1 and MYH3) are interpreted as carrier findings rather than disease‐causing in these probands.

3.6. MAE Analysis

Integration of DNA methylation with RNA sequencing data from 10 trios enabled assessment of MAE patterns. Of 5.6 million informative heterozygous SNPs with determinable parental origin, 223,753 (4.7%) had sufficient RNA coverage for allele‐specific expression analysis. The overall MAE rate was 0.3% (656 sites), within the expected normal range, with no family exceeding 1%, indicating no evidence of widespread allele‐specific silencing or imprinting defects.

3.7. Epigenetic Age Acceleration

Analysis of epigenetic age revealed differences between patient groups (Figure 1). Both the Hannum and Horvath clocks showed a large biological age acceleration in HLHS patients compared with their chronological age. This acceleration was significantly greater than both the NDD‐only group (mean: 9.5 years greater, p = 0.006) and control subjects (mean: 7.4 years greater, p = 0.008), even after accounting for relatedness using mixed‐effects models. The NDD‐only probands showed moderate age acceleration that was not significantly different from controls (p = 0.31). Interestingly, when stratified by family role, probands across all disease categories showed higher age acceleration than their parents (probands: mean: 11.1 years; mothers: mean: 6.8 years; fathers: mean: 6.6 years).

Figure 1: Epigenetic age acceleration in trios with CHD and/or NDD. Hannum age acceleration stratified by family relationship and disease status. Box plots showing the distribution of age acceleration (epigenetic age minus chronological age) calculated using the Hannum blood‐specific clock. The cardiac group (HLHS patients, n = 3) shows markedly higher age acceleration compared with control subjects (n = 12) and NDD‐only patients (n = 21). Each point represents an individual sample. Boxes indicate interquartile range (IQR), centre lines show medians and whiskers extend to 1.5 × IQR.

A second analysis comparing the probands and healthy controls also revealed elevated biological aging in CHD probands (Figure S4). These alternate epigenetic clocks, EN and Levine, showed similar epigenetic age results for the CHD probands, but the EN clock had much higher ages for controls and NDD probands. Although both clocks confirmed a higher epigenetic age in CHD probands compared with NDD probands and controls, this result was not statistically significant.

3.8. Differential DNA Methylation Analysis

Cross‐cohort differential methylation analysis comparing individual probands to controls (n = 178) revealed methylation dysregulation in four probands (Figure 2). Proband 13 showed the most severe dysregulation with 72,355 DMPs, followed by Proband 19 (56,816 DMPs), Proband 20 (49,618 DMPs) and Proband 15 (10,037 DMPs). All remaining probands had fewer than 10,000 DMPs.

Figure 2: Differentially methylated positions in probands compared with controls. Individual proband methylation analysis identified four outlier cases with extreme methylation dysregulation (Probands 13, 15, 19 and 20) showing > 10,000 DMPs, whereas the remaining probands displayed moderate changes (149–6568 DMPs). Genomic context distribution shows distinct patterns across outlier probands: (A) Proband 13 (72,355 total DMPs) shows predominant hypomethylation in open sea regions (23,602 sites) with balanced hypermethylation in CpG islands (19,275 sites). (B) Proband 15 (10,037 DMPs) exhibits predominantly hypermethylation in CpG islands (2387 sites) and open sea regions (3126 sites). (C) Proband 19 (56,816 DMPs) displays extensive hypomethylation in open sea regions (25,483 sites) with hypermethylation in shores (10,392 sites). (D) Proband 20 (49,618 DMPs) shows a similar pattern with 23,602 hypomethylated open sea sites. DMPs were defined using limma comparison against 178 control samples (FDR < 0.05, |Δ β| > 0.2). Genomic contexts defined per Illumina MethylationEPIC array annotation. Definitions are as follows: CpG islands are regions with > 50% GC content, > 60% CpG observed/expected ratio and minimum length of 200 bp; shores are regions flanking CpG islands within 2 kb (2000 bp); shelves are regions 2–4 kb (2000–4000 bp) from CpG islands; and open sea regions are isolated CpGs located > 4 kb (4000 bp) from any CpG island.

The genomic distribution of DMPs varied substantially across probands (Figure 2). In Proband 13, hypermethylated sites were enriched in CpG islands (19,275 sites, 76.2% of CpG island DMPs) whereas hypomethylated sites were predominantly in open sea regions (21,406 sites, 86.6% of open sea DMPs), indicating disruption of normal methylation maintenance mechanisms rather than targeted changes at specific loci. Probands 19 and 20 had similar methylation patterns across genomic contexts, with greater levels of hypomethylation in CpG islands and markedly greater levels of hypomethylation across open seas, shelves and shores. Finally, Proband 15 had quite a different pattern of methylation dysregulation, showing higher levels of hypermethylation across all genomic contexts.

3.9. Pathway Enrichment Analysis of DNA Methylation Regions

DMRs were enriched for pathways related to both cardiac and neurological processes (Table S4). Key pathways included muscle tissue development, muscle cell differentiation and striated muscle cell differentiation, which align with the structural and functional roles of muscle tissues in cardiac development. Pathways such as regulation of blood circulation and regulation of heart contraction further highlight the involvement of these DMRs in cardiac‐specific biological processes. Additionally, enrichment in pathways related to neuronal development, such as axon guidance, neuron projection guidance, synapse organisation and transmission of nerve impulse, suggests a role for these epigenetic changes in nervous system function. Processes like stem cell differentiation, cell fate commitment and pattern specification process show broader developmental dysregulation.